Endocrine Abstracts

Background: Bulk genomic studies have identified distinct molecular classes of adrenocortical tumors (ACT). Transcriptome profiles separate benign ACT (“C2” cluster) from carcinomas (ACC) and identify two groups of ACC, “C1A” (“steroid” and “proliferation” signatures) and “C1B” (“immune” signature), of poor and better prognosis respectively. However, these signatures were characterized at the tissue level (“bulk”) and our knowledge of the cell composition of ACT is limited.

Aim: This study aims at providing a single-nucleus atlas of human normal adrenal cortex and ACT.

Methods: We performed single-nucleus RNA-sequencing (10×) of ~170,000 cells from normal adrenal (n=4), benign ACT (n=14) and ACC (n=20). Cell Ranger and Seurat pipelines were used for deciphering the cellular heterogeneity of steroid and microenvironment cells. Steroid and microenvironment signatures were then scored (ssGSEA and CIBERSORTx) in “bulk” transcriptomes of 201 ACC patients (ENSAT 2014 (n=33), TCGA 2016 (n=79) and ENSAT 2022 (n=89)), to define tumor ecotypes (consensus clustering) and to test their association with outcome.

Results: In steroid tumor cells, 8 variable expression programs were identified, including 4 related to steroidogenesis, mirroring normal adrenal functional zonation, and 4 related to oncogenic processes. Microenvironment of steroid cells was scarce, mainly composed of fibroblasts (4.7% of total cells), endothelial cells (5.1%), macrophages (9.9%) and T cells (1.2%). ACC microenvironment was characterized by cancer-associated signatures, including cancer-associated fibroblasts (CAF, expressing PDGFRB and FN1 ), tumor-associated endothelial cells (expressing ANGPT2 and VWF ) and tumor-associated macrophages (expressing CD163 and F13A1 ). Compared to “C1A”, “C1B” ACC were enriched in inflammatory “M1-like” macrophages. Deconvolution in ACC bulk transcriptomes identified 3 tumor ecotypes associated with outcome. A first ecotype combined cancer-associated fibroblasts and tumor-associated endothelial cells, with hypoxia and mitosis signatures in steroid cells. A second ecotype combined exhausted T cells, with zona fasciculata steroid signature. These ecotypes were associated with “C1A” and poor survival (log-rank P<10^-6 and <10^-9 respectively). Conversely, a third ecotype combined inflammatory macrophages, with zona reticularis steroid signature, and was associated with “C1B” and better outcome (log-rank P<10^-7 ).

Conclusion: Single-cell steroid and microenvironment signatures combine into tumor ecotypes associated with outcome, participating in the understanding of ACC heterogeneity and the C1A/C1B partition that we previously reported.